- Email: [email protected]
Narrative language in traumatic brain injury
Neuropsychologia 49 (2011) 2904–2910
Contents lists available at ScienceDirect
Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia
Narrative language in traumatic brain injury Andrea Marini a,b,∗ , Valentina Galetto c,d , Elisa Zampieri b , Lorenza Vorano e , Marina Zettin c,d , Sergio Carlomagno f a
Dipartimento di Scienze Umane, University of Udine, Udine, Italy IRCCS “E. Medea: La Nostra Famiglia”, San Vito al Tagliamento (Pn), Italy Centro Puzzle, Torino, Italy d Dipartimento di Psicologia, Università degli Studi di Torino, Italy e IMFR Gervasutta, Udine, Italy f Dipartimento di Psicologia, Università di Trieste, Italy b c
a r t i c l e
i n f o
Article history: Received 9 November 2010 Received in revised form 14 June 2011 Accepted 15 June 2011 Available online 22 June 2011 Keywords: Traumatic brain injury Narrative analysis Neurolinguistics Neuropsychology
a b s t r a c t Persons with traumatic brain injury (TBI) often show impaired linguistic and/or narrative abilities. The present study aimed to document the features of narrative discourse impairment in a group of adults with TBI. 14 severe TBI non-aphasic speakers (GCS < 8) in the phase of neurological stability and 14 neurologically intact participants were recruited for the experiment. Their cognitive, linguistic and narrative skills were thoroughly assessed. The group of non-aphasic individuals with TBI had normal lexical and grammatical skills. However, they produced narratives with increased errors of cohesion and coherence due to the frequent interruption of ongoing utterances, derailments and extraneous utterances that made their discourse vague and ambiguous. They produced a normal amount of thematic units (i.e. concepts) in their narratives. However, this information was not correctly organized at micro- and macrolinguistic levels of processing. A Principal Component Analysis showed that a single factor accounted for the production of global coherence errors, and the reduction of both propositional density at the utterance level and proportion of words that conveyed information. It is hypothesized that the linguistic deficits observed in the participants with TBI may reflect a deficit at the interface between cognitive and linguistic processing rather than a specific linguistic disturbance. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Individuals with severe traumatic brain injury (TBI) often show a variety of communicative difficulties. Indeed, problems in dealing with non-literal expressions (i.e. sarcasm, humor, proverbs, idioms, and indirect requests) have been frequently reported (Angeleri et al., 2008; Docking, Murdoch, & Jordan, 2000Docking, Murdoch, and Jordan, 2000; McDonald & Pearce, 1996). Furthermore, their discourse has been described as inefficient, impoverished or even confused (Davis & Coehlo, 2004; Hartley & Jensen, 1991, 1992). In the analysis of their linguistic performance an important distinction has been introduced between microlinguistic and macrolinguistic levels of discourse processing (e.g. Glosser & Deser, 1990). Microlinguistic abilities refer to within-sentence construction (i.e. lexical and morpho-syntactic aspects of language processing) analyzed in terms of lexical errors (e.g. phonological errors, verbal or semantic paraphasias, or use
∗ Corresponding author at: Università di Udine, Via Margreth, 3, 33100 Udine, Italy. Tel.: +39 335 5393224. E-mail address: [email protected] (A. Marini). 0028-3932/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.neuropsychologia.2011.06.017
of indefinite terms) and/or syntactic organization (e.g. proportion of complete sentences, syntactic complexity) (Caplan, 1992). Macrolinguistic abilities relate to pragmatic and discourse-level aspects of language processing, i.e. those recruited in establishing cohesive and conceptual links among contiguous (cohesion and local coherence) or long-distant sentences/utterances (global coherence) and in formulating a mental model or gist of a story or procedure. Accordingly, macrolinguistic measures include the analysis of errors of cohesion, local and global coherence (e.g. presence of tangential utterances or extraneous propositional content). Several studies have shown potential dissociations between microlinguistic and macrolinguistic deficits in individuals with TBI. Nonetheless, Glosser and Deser (1990) reported that nine participants with TBI were impaired in lexical, syntactic and suprasentential organization. However, these persons were diagnosed with a form of fluent aphasia that may have hampered their microlinguistic processing. Different results come from other studies (e.g. Mentis & Prutting, 1987; Coelho, 2002). Indeed, when compared to a group of healthy participants, three individuals with TBI produced fewer cohesive ties in procedural descriptions but showed relatively preserved syntactic skills (Mentis & Prutting,
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
1987). In a similar vein, Coelho (2002) studied a cohort of 55 consecutive participants with TBI who were not aphasic on a story telling task. These individuals did not differ from a group of healthy participants in terms of sentence complexity and cohesive adequacy (i.e. proportion of complete cohesive ties) but introduced in their narratives more extraneous propositional content, suggesting problems in the organization of information at the between-sentence level. This shows that even cohesion and coherence may be dissociable components of discourse processing (see on this point Glosser, 1993). The reduced informative content in their discourse might contribute to the perception of impoverished and confused language. However, Ehrlich (1988) examined the Cookie Theft Picture descriptions from 10 non-aphasic adults with TBI with the Content Units method (CUs, Yorkston & Beukelman, 1980) and found them normal in the amount of CUs but inefficient. They uttered fewer CUs per minute than healthy participants. More recently, similar results have been reported by Stout, Yorkston, and Pimental (2000) and Carlomagno, Giannotti, Vorano, and Marini (2008) who analyzed the story-narratives from 94 and 10 non-aphasic participants with TBI, respectively. In the latter study, the persons with TBI showed normal microlinguistic processing but were found impaired on macrolinguistic measures. Interestingly, the occurrence of macrolinguistic errors correlated with the rating of language inaccuracy by naïve judges. These results further suggest that the impression of confused and impoverished language from non-aphasic individuals with TBI may depend on the reduced ability to organize information at the macrolinguistic level of processing rather than on difficulties in dealing with lexical and syntactic (i.e. microlinguistic) aspects of language production. The possibility of a problem in the global organization of information at the text level is supported by findings from studies focusing on story grammars. These refer to “the internal structure of stories which guide an individual’s comprehension and production of logical relationships, both temporal and causal, between agents and events” (Cannizzaro & Coelho, 2002, p. 1065). As such, story grammars rely on the interaction between language, memory and executive functions. Importantly, the ability to structure a story in terms of complete episodes seems altered in individuals with TBI (e.g. Brookshire, Chapman, Song, & Levin, 2000; Chapman et al., 1992; Davis & Coehlo, 2004). As for the distribution of information at the microlinguistic level, Coelho, Grela, Corso, Gamble, and Feinn (2005) analyzed the ability of a cohort of participants with TBI to insert multiple ideas into single sentences, i.e. evaluating the propositional density of their sentences. When compared to a group of healthy participants, the group of individuals with TBI produced significantly fewer propositions per T-unit (i.e. sentence). According to the authors, this suggests impairments of both micro- and macrolinguistic processes involved in the organization of semantic information in discourse. Furthermore, they proposed that such impairments “are symptomatic of generalized cognitive disruptions as opposed to specific linguistic deficits” (p. 1144). More recently, however, Ellis and Peach (2009) have examined the pausing patterns of TBI and healthy participants who had been administered sentence production (i.e. repetition and reading) tasks. They found increased pausing in the group of participants with TBI. Interestingly, their pausing pattern strongly correlated with a measure of sentence complexity. This led the authors to conclude that they may have specific deficits with online sentence processing even when normal on measures of syntactic complexity. Notably, in a preceding study by Peach and Schaude (1986) (quoted by Ellis & Peach, 2009), the narratives produced by 20 TBI patients were found normal on measures of syntactic complexity but contained more syntactic errors including word-order transposition, verb tense and agreement errors (see also Glosser & Deser, 1990).
2905
Overall, two major aspects of narrative discourse in non-aphasic individuals with TBI should be taken into account when interpreting the nature of their confused language. First, they seem to produce a normal amount of information content in their narratives although at reduced rate (Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). Second, this information may be poorly organized at both between-sentence level, as shown by the production of global coherence errors (Coelho, 2002; Carlomagno et al., 2008), and within-sentence level, as the results by Coelho et al. (2005) seem to indicate. However, a relevant question here is whether the microlinguistic deficit in the narratives produced by persons with TBI is due to specific linguistic problems (e.g. Ellis & Peach, 2009; Peach & Schaude, 1986) or to more generalized cognitive dysfunctions (e.g. Coelho et al., 2005). The present study further investigates the narrative skills of a group of non-aphasic individuals with TBI by analyzing in detail their narratives elicited in a single-picture and cartoon story description task. It was hypothesized that, in spite of preserved microlinguistic abilities and normal performance in terms of information content, these individuals would exhibit impaired organization of information at textual level producing several violations of global coherence. Furthermore, we expected to replicate the finding by Coelho et al. (2005) of reduced semantic content at the sentential level. Finally, we examined the relationship between linguistic and informative aspects of language production in TBI by performing a Principal Component Analysis. 2. Materials and methods 2.1. Participants Twenty-eight Italian-speaking participants were included in the study. They formed two groups balanced for age and level of formal education (Table 1). The experimental group was formed by 14 individuals with traumatic brain injury (TBI) (age – mean: 35.4; SD: 8.5; level of formal education – mean: 10.9; SD: 2.6). Their mean score on the Glasgow Coma Scale in the acute phase had been less than 8 (days of coma – mean: 32.4; SD: 18.4; range: 5–59) (see Table 1). In order to be included in the experimental group the patients had to be in the phase of neurological stability (months post-onset – mean: 68.5; SD: 38; range: 15–134) and normal to near-normal performance on the Aachener Aphasie Test (AAT, Italian version, Luzzatti, Willems, & DeBleser, 1991). The control group was formed by 14 neurologically healthy controls (HC) (Age – mean: 35.5; SD: 6.1; Level of formal education: 12.3; SD: 1.8) (Table 1). Inclusion criteria for admission in the control group included normal range performance on Raven’s progressive matrices (Raven, 1938) and normal performance on a series of neuropsychological tests (see Section 2.2.1 and Table 2). None of the participants had a previous history of psychiatric or neurological illness, learning disabilities nor hearing or visual loss. All participants released their written informed consent to participate to the study after all procedures had been fully explained. Approval for the study had previously been obtained from the local ethic committee. 2.2. Procedures 2.2.1. Neuropsychological assessment The cognitive profile of all participants was carefully evaluated. The neuropsychological assessment focused on those cognitive functions which are needed to complete the narrative task, i.e. phonological and semantic verbal fluency, phonological short-term memory (digit span), verbal learning (Rey’s 15-word Immediate Recall and Delayed Recall), executive functioning (Wisconsin Card Sorting Test, WCST perseverative and non-perseverative errors; Heaton, Chelune, Talley, Kay, &
Table 1 Means (and standard deviations) of demographic and clinical characteristics of the groups of traumatic brain injured (TBI) and healthy control (HC) participants. TBI
Age Formal education (years) Time after injury (months) Coma duration (days) GCS (score)
HC
Mean (SD)
(Range)
35.4 10.9 68.5 32.4 4.9
(18–50) 35.5 (8–13) 12.3 (15–134) (5–59) (3–8)
(8.5) (2.6) (38) (18.4) (1.7)
Mean
(SD) (Range) (6.1) (20–44) (1.8) (8–13) – – –
2906
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
Table 2 Means (and standard deviations) of the neuropsychological performance of the groups of traumatic brain injured (TBI) and healthy control (HC) participants.
TMT-A (s)* TMT-B (s)* Phonemic fluency Semantic fluency* WCST (pers. err.)* WCST (non-pers. err.)* Rey’s 15-word immediate recall* Rey’s 15-word delayed recall* Phonological short-term memory (digit span – forward)* Phonological short-term memory (digit span – backward)*
TBI
HC
82.6 (39.2) 198.6 (73.8) 24.1 (11.1) 12 (3.2) 47.4 (34.4) 18.4 (13.7) 27.7 (10) 3.1 (4.3) 5.1 (.9)
52.1 (11.1) 97.1 (28) 32.9 (8.6) 23.1 (3.9) .1 (.3) .2 (.4) 45.9 (6.7) 9.6 (1.9) 7 (1.1)
3.8 (.9)
5.6 (1.2)
GCS: Glasgow Coma Scale; TMT-A: Trail Making Test, Part A; TMT-B: Trail Making Test, Part B; WCST (pers. err): Wisconsin Card Sorting Test, perseverative errors; WCST (non-pers. err): Wisconsin Card Sorting Test, non-perseverative errors. * Significant group-related difference. The score on the phonological short-term memory was not available for the HC group.
Curtiss, 1993) and sustained and selective attention (Trail Making Test, Parts A and B, respectively; Reitan, 1992). Neuropsychological scores are shown in Table 2. 2.2.2. Assessment of narrative abilities The assessment of narrative abilities was performed using story telling. Each participant was asked to produce a set of narratives elicited with the help of one single picture depicting a shot of a story (the scene of a “Picnic”) and two cartoon stories with six pictures each presented on the same page (the stories of the “Flower Pot” and of a “Quarrel”). The single picture “Picnic” was taken from the Western Aphasia Battery (WAB; Kertesz, 1982). The two cartoon picture sequences have been used by Huber and Gleber (1982) and Nicholas and Brookshire (1993), respectively, for analyzing textual competence and discourse information content in brain damaged subjects. Each subject told all three stories. The order of presentation was counterbalanced across subjects. In order to avoid poor performance due to short-term memory limitations, the picture or cartoon story remained visible until the subject had finished his/her description. Each storytelling was tape-recorded and subsequently transcribed verbatim by three of the authors (AM, LV, VG); the transcription included phonological fillers, pauses, false starts and extraneous utterances. These transcriptions underwent quantitative, in-depth linguistic and textual analysis focusing on five main aspects of linguistic processing: productivity, speech rate, lexical and syntactic encoding, textual organization and informativeness (Marini, Andreetta, del Tin, & Carlomagno, 2011). Both simple values and ratios were considered to account for the difference across the narratives. An example of transcriptions and analyses of narratives from persons with TBI please is provided in Appendix A. Productivity measures included units, words, speech rate and mean length of utterance (MLU). The unit count included each word, non-word or syllabic false start uttered by the speaker. The total number of well-formed words with the exception of phonological fillers, phonemic paraphasias and phonetic errors was then computed. The number of words was used to obtain a measure of speech rate in terms of words per minute (Words/m ). For each story description, the total number of utterances was assessed following the criteria established in the Shewan Spontaneous Language Analysis System (Shewan, 1988). MLU was calculated by dividing the total number of words by the number of utterances. Lexical processing was assessed in a two-step analysis. At step one the analysis concerned the lexeme level of word processing (Levelt, Roelofs, & Meyer, 1999). The participants’ ability to retrieve phonologically well-formed words was assessed computing an index of phonological selection. This was obtained by dividing the number of words by the number of units (Marini, Carlomagno, Caltagirone, & Nocentini, 2005a). At step two the lemma level was assessed in terms of proportion of semantic and verbal paraphasias or paragrammatic errors in the sample. The former index was calculated by dividing the number of semantic and verbal paraphasias by the number of words (Haravon, Obler, & Sarno, 1994). The index of paragrammatic errors was calculated by dividing the number of paragrammatic errors by the number of words. These errors included misuse of bound morphemes and function words. The identification of grammatically complete sentences provided a measure of grammatical organization. An index was calculated by dividing the number of grammatical sentences by the number of utterances (Saffran, Berndt, & Schwartz, 1989; Thompson, Shapiro, Tait, Jacobs, & Schneider, 1996). A sentence was considered grammatically complete if all the arguments required by the verb were inserted correctly in the body of the sentence and if there were no omissions or substitutions of free or bound morphemes. As for discourse organization at intersentential level, the macrolinguistic measures included indexes of cohesion and global coherence errors. The index of
cohesive errors measured the extent to which each utterance of the narrative was structurally related to the preceding one. Cohesion errors were misuse of cohesive ties like anaphoric pronouns, errors in number and gender agreement between pronouns or noun phrases across utterances, misuse of either cohesive functionwords or semantically related content words, and abrupt interruptions of utterances (Haravon et al., 1994). The occurrence of cohesive errors was evaluated by dividing the total number of cohesive errors by the number of utterances produced (Marini, Boewe, Caltagirone, & Carlomagno, 2005b). With regard to global coherence, a ratio of global coherence errors was calculated by dividing the total number of tangential and extraneous utterances by the number of utterances that formed each description. The information content of each narrative was evaluated at three levels. The first one concerned the number of thematic units contained in the speech sample. A thematic unit was defined as a main idea or detail in the story that has been identified in a previous study (Marini, Carlomagno, et al., 2005). The number of thematic units produced by each subject was considered an index of the participants’ abilities to derive information from the picture stimuli. The second measure concerned the production of appropriate lexical information units (LIUs), i.e. those content and function words that were not only phonologically well-formed but also appropriate from a grammatical and pragmatic point of view (Marini, Boewe, et al., 2005; Marini, Carlomagno, et al., 2005). Therefore, all those words that were classified as semantic or verbal paraphasias, fillers, paragrammatic errors or forming tangential or extraneous utterances (i.e. utterances that were somehow deviating from the gist of the story) were excluded from the LIUs’ count. An index of lexical informativeness was then obtained by dividing the number of LIUs by the number of words. This ratio (%LIUs) has proved adequate in measuring effectiveness in encoding information in speech in several neurologically impaired populations and neuropsychiatric conditions (e.g. right brain damaged subjects, Marini, Carlomagno, et al., 2005; schizophrenic patients, Marini et al., 2008). Furthermore, in order to assess the participants’ ability to insert multiple ideas into single utterances, an index of “Thematic Density” was computed by dividing the number of thematic units produced by each participant for each story by the number of utterances that made up the speech-samples. This measure roughly reproduced the propositional complexity index used by Coelho et al. (2005) allowing us to examine the semantic complexity of utterances apart from sentence structure and grammaticality. The scoring procedure was performed independently by two raters and then compared. Acceptable interrater reliability was defined as k ≥ 0.80. After an appropriate training, the two raters, achieved this level prior to the beginning of the study. The residual differences were resolved through discussion. 2.3. Statistical analyses Group related differences in neuropsychological scores were assessed entering a Group (1. TBI; 2. C) X Task (phonemic and semantic fluency, TM-A, TM-B, and Rey’s 15-words immediate and delayed recall) MANOVA. As for the group-related differences on perseverative and non-perseverative errors on the WCST, a Mann–Whitney U test was performed. The narrative performance of the two groups of participants was analyzed performing a two-way ANOVA with repeated measures with group as between-subject factor and story as within-subject factor for each sample on twelve measures (words; speech rate; mean length of utterance; % phonological selection; % semantic/verbal paraphasias; % paragrammatic errors; % complete sentences; % cohesion errors; % global coherence errors; number of thematic units; % lexical information units; index of thematic density). The level of statistical significance was set at p < .004 after Bonferroni correction for muplitple comparisons. When significant, the story*group interaction will be reported. A correlation analysis was performed between neuropsychological and narrative scores in the group of participants with TBI. Finally, a Principal Component Analysis with Orthotran Varimax rotation was performed to identify factors explaining measures of informativeness and macrolinguistic processing.
3. Results 3.1. Cognitive and neuropsychological assessment Neuropsychological scores are shown in Table 2. The performance of the TBI group was significantly different from that of the group of healthy participants on all measures with the only exception of phonemic fluency. Namely, with respect to the control group they were slower on both TM-A ([F(1; 22) = 7.845; p = .010]) and TM-B ([F(1; 22) = 23.246; p = .000]) and they produced less words on the Semantic Fluency task ([F(1; 22) = 49.669; p = .000]). Furthermore, they recalled fewer items on Rey’s 15-word immediate ([F(1; 22) = 26.802; p = .000]) and delayed recall ([F(1; 22) = 22.595; p = .000]), respectively. The group-related difference
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
was also significant on the Digit Span test where the TBI group performed worse than the HC group on both forward and backward digit span ([F(1; 22) = 21.160; p = .000] and [F(1; 22) = 19.151; p = .000], respectively). As for the production of perseverative and non-perseverative errors on the WCST, the difference was significant in both cases: perseverative errors (Z = −4.498; p = .000); non-perseverative errors (Z = −4498; p = .000).
2907
3.2.3. Analysis of informativeness The mean values for each group on the three measures of informativeness are reported in Table 4. The group-related differences were not significant on thematic informativeness, i.e. the speech samples by the TBI subjects contained the same number of main ideas and details as the narratives from the healthy controls. However, the proportion of words that were counted as Lexical Information Units was reduced with respect to healthy controls ([F(1; 26) = 36.710; p = .000; partial 2 = .585]) indicating that relevant information was mixed with a number of unnecessary words, tangential speech or repetitions. Furthermore, the TBI participants also produced a significantly lower number of thematic units per utterance ([F(1; 26) = 35.251; p = .000; partial 2 = .576]). On the whole, the results of the information content analysis confirmed that narrative speech from non-aphasic TBI subjects contained a normal amount of information (see also Carlomagno et al., 2008; Ehrlich, 1988). This, however, was produced less efficiently as indicated by the reduced proportion of informative words (Carlomagno et al., 2008) and reduced semantic content per utterance.
3.2. Assessment of narrative abilities The results will be presented in three separate sections: a Microlinguistic analysis section, a Macrolinguistic analysis section and an Informative analysis section.
3.2.1. Microlinguistic analysis The mean values for each group on each microlinguistic measure are reported in Table 3. The two groups produced descriptions with a comparable number of words and MLU. However, the group of TBI participants produced narratives with significantly slower speech rate than the control group ([F(1; 26 = 13.684; p < .001; partial 2 = .345]). Both groups formulated descriptions with a similar index of phonological selection, semantic/verbal paraphasias and complete sentences. As for the production of paragrammatic errors, the group-related difference was significant ([F(1; 26) = 14.953; p < .001; partial 2 = .365]). However, it is noteworthy that the paragrammatic errors produced by the TBI participants were in any case very limited (see Table 3).
3.3. Correlations between neuropsychological scores and narrative performance in TBI participants A series of correlations were performed between those neuropsychological variable scores (i.e. non-perseverative and perseverative errors on the Wisconsin Card Sorting Test, Rey’s 15-word immediate recall and Rey’s 15-word delayed recall, Semantic Fluency, TMT-A and TMT-B) and those measures of the narrative production that were found altered in the TBI patients (i.e. speech rate, % cohesive errors, % global coherence errors, % lexical informativeness and the ratio of thematic density). The results failed to show any significant correlation between these scores.
3.2.2. Macrolinguistic analysis The mean values for each group on each macrolinguistic measure are reported in Table 4 together with data from the information content analysis. The TBI participants produced significantly more violations of both cohesion (F(1; 26 = 12.888); p < .001; partial 2 = .331]) and global coherence ([F(1; 26) = 31.134; p = .000; partial 2 = .553]) rules. However, the cohesion errors also included abrupt interruption of utterances (Haravon et al., 1994). When these errors were discarded from the computation, the other cohesive errors (e.g. misuse of anaphoric pronouns, errors in number and gender agreement between pronouns or noun phrases across utterances, misuse of either cohesive function-words or semantically related content words) did not differ in the two groups.
3.4. Principal component analysis of measures of informativeness and macrolinguistic accuracy In order to examine if separate factors corresponding to hypothesized components of micro- and macrolinguistic processing abilities could be empirically identified, data of each sample concerning measures of informativeness (% lexical information units and ratio of thematic density) entered a principal component analysis with Orthotran Varimax rotation together with the main measures of impaired macrolinguistic processing (% cohesion
Table 3 Results of the microlinguistic analysis for the groups of TBI and healthy control participants. Microlinguistic analysis
TBI
HC
Level of significance (p)
Effect size (partial 2 )
Words Speech rate* MLU % Phonological selection % Semantic paraphasias % Paragrammatic errors* % Complete sentences
82.5 (31.3) 94.7 (29.7) 5.6 (1.1) 99.2 (1) .8 (1.1) 1.4 (1.1) 57.9 (15.3)
80.9 (44.4) 129.5 (29.3) 6.9 (1.9) 99.6 (.8) .1 (.4) .2 (.6) 63.1 (23.1)
<.964 <.001 <.013 <.206 <.024 <.001 <.412
.000 .345 .213 .061 .181 .365 .026
*
When the group-related difference is significant after Bonferroni correction for multiple comparisons.
Table 4 Results of the analysis of the macrolinguistic and informative aspects of narrative production for the groups of TBI and healthy control participants. Macrolinguistic and informative analysis *
% Cohesive errors % Global coherence errors* % Lexical informativeness* Thematic informativeness Ratio of thematic density* *
TBI
HC
Level of significance (p)
Effect size (partial 2 )
3.9 (1.9) 22.1 (11.1) 64.3 (10.7) 6 (2) .4 (.2)
1.9 (.4) 3.8 (7.5) 84.5 (9.6) 7 (2) 1.2 (.6)
<. 001 .000 .000 <.037 .000
.331 .553 .585 .156 .576
When the group-related difference is significant after Bonferroni correction for multiple comparisons.
2908
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
errors and % global coherence errors) and speech rate. The latter was the only measure that could be assumed as a marker of microlinguistic deficit in the TBI group. Indeed, as mentioned above the % paragrammatic errors was higher in the participants with TBI but unlikely the small amount of these errors could explain the main features of their narrative speech. The principal component analysis showed that two factors could explain approximately 78% of variance (Bartlett’s 2 = 94.2, p < .0001). The first one, accounting for 50.5% of variance, included measures of % global coherence errors (−.93), % lexical information units (.93), and the ratio of thematic density (.83). The second factor, accounting for 27.5%, included the % cohesion errors (.80) and speech rate (−.79).
4. Discussion This study investigated cognitive, linguistic and narrative abilities of a group of severe TBI non-aphasic speakers while in the phase of neurological stability compared to a group of healthy participants. The neuropsychological assessment showed persisting difficulties in several areas of the cognitive domain. The analysis of narratives included microlinguistic, macrolinguistic and informative aspects of narrative language elicited in a cartoon-story description task. The main results indicate a significant deficit on the macrolinguistic and informative dimensions of processing in absence of important deficits of language processing on the withinsentence level. When evaluated on measures of verbal productivity and grammatical and lexical adequacy, the story narratives produced by the individuals with TBI were adequate. They had normal length (as measured in terms of uttered words and mean length of utterance), and showed no sign of phonological, morphological, semantic or syntactic deficits. Indeed, the only significant group-related difference concerned the production of paragrammatic errors and speech rate. The former, however, was limited to barely 1% of uttered words. Such a modest morpho-syntactic deficit unlikely accounts for the impoverished and confused language found in these patients. Interestingly, their narrative descriptions were produced with slower rate. Speech rate can be affected by several factors such as phonetic/phonological impairments, reduced lexical selection abilities or even macrolinguistic deficits. In our study, however, phonetic/phonological deficits can be ruled out, as the participants with TBI had normal phonological selection abilities. Similarly, problems in lexical access do not seem likely as shown by the absence of verbal or semantic paraphasias in their narratives. It remains open the possibility of a connection with macrolinguistic deficits. Indeed, the data discussed here showed reduced macrolinguistic competence in establishing the intersentential connections typical of a narrative structure. The narratives from the individuals with TBI were characterized by frequent interruptions of ongoing utterances (e.g. / the man is staring at . . . / the man is watching the dog /), and derailments (e.g. / It is a picnic / I like picnics / I have made several picninc in my life /) that made their discourse vague and ambiguous. The Principal Component Analysis showed that these macrolinguistic errors accounted not only for the reduced levels of cohesion and global coherence, but also for the reduced levels of speech rate and informativeness. Speech rate and Cohesion errors were explained by one single factor. Importantly, our evaluation of cohesion errors included the abrupt interruptions of ongoing utterances (Haravon et al., 1994). When interruption errors were discarded from the computation, the occurrence of cohesion errors, i.e. misuse of cohesive ties, was not different across the two groups. Therefore, it is likely that the reduced speech rate may have been determined by the frequent interruptions in the flow of thoughts rather than to specific linguistic deficits.
Another interesting finding concerns the organization of information in the narratives. The group of participants with TBI produced a normal amount of thematic units. This suggests that they could identify and introduce all expected thematic units (see also Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). However, they also exhibited reduced performance on two functional measures of informativeness. First, they produced a lower proportion of words that were scored as lexical information units (%LIUs). Second, their narratives contained fewer thematic units per utterance. The reduced %LIUs reflects a macrolinguistic deficit in coherently linking concepts in a story format by controlling what has been said, integrating the latter with incoming information, and inhibiting irrelevant behaviors or the production of tangential utterances. Indeed, words embedded in tangential utterances or forming extraneous propositional content were excluded from the LIUs count. However, the utterances produced by the participants with TBI had also less thematic density than those produced by the group of healthy participants. This finding is similar to that reported by Coelho et al. (2005) who interpreted it as a sign of generalized cognitive deficit influencing the organization of the semantic content of sentences. As mentioned in the Introduction section, Ellis and Peach (2009) suggested that the linguistic impairment found in persons with TBI may reflect specific linguistic deficits in online sentence planning. However, our data indicate that the participants TBI did not have relevant problems at the sentential level. Indeed, they produced the same amount of grammatically complete sentences as the healthy participants. Furthermore, a deficit in sentence planning could unlikely account for the production of errors of global coherence. Interestingly, in our study the principal component analysis indicated that a single factor accounted for the reduced %lexical information units, the reduced index of thematic density and the high % of global coherence errors. Taken together, all these symptoms may reflect a general cognitive problem in organizing the semantic content of the story rather than a specific linguistic deficit in coherently linking the propositions that form the narrative or in planning its sentences. This interpretation of the data is further supported by the neuropsychological profile of the participants. Indeed, the neuropsychological assessment showed important difficulties in several areas of the cognitive domain such as the ability to access lexical information stored in the mental lexicon by means of a semantic strategy (semantic verbal fluency), verbal learning skills (Rey’s 15-word Immediate Recall and Delayed Recall), short-term memory (forward and backward digit span), sustained and selective attention abilities (TM-A and TM-B), and executive functioning (perseverative and non-perseverative errors on the Wisconsin Card Sorting Test and phonemic fluency). The impairment in attention is among the most frequently reported sequelae of TBI (Stierwalt & Murray, 2002). This is a pervasive deficit that might affect several mental processes therefore contributing to other cognitive problems. As such, it may be involved in the observed difficulty in encoding and storing items to memory, in generating solution to problems and in language processing. In a similar vein, the anomalous performance on the WCST suggests that the participants with TBI may not be able to adequately select new strategies for solving problems and inhibit inappropriate responses. However, in our study none of these neuropsychological scores correlated with any of the linguistic measures mentioned so far (i.e. % lexical information units, index of thematic density, % global coherence errors). Even if this lack of correlation between linguistic and cognitive measures has been frequently reported in TBI, some studies have highlighted the potential relationship between executive functioning, memory and higher-level linguistic processing (e.g. Snow & Douglas, 2000; Snow, Douglas, & Ponsford, 1998). This apparent discrepancy can be explained in terms of a global cognitive impairment that becomes explicit when these patients fail to correctly organize the ideas they
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
intend to convey in a narrative. This poor organization of information may transcend the cognitive level and affect both macro- and microlinguistic processing even when the amount of information is normal. In other words, in a picture story description task as the one used in this experiment, the patient may extract all necessary information about the story but is then unable to organize it at both microlinguistic (as shown by the lowered propositional density of his/her utterances) and macrolinguistic (low %LIUs) levels. Such problems in the organization of information may therefore explain also the high production of macrolinguistic errors in terms of a compensatory strategy. Indeed, facing with this massive problem in organizing the information extracted by the pictures, the patient likely tries to express all these ideas randomly, without a clear organization. We therefore hypothesize that all these microand macrolinguistic symptoms may reflect a deficit in the interface between cognitive and linguistic processing as hypothesized by Glosser (1993). A further point that deserves discussion concerns the apparent discrepancy between the absence of linguistic deficits on the general linguistic assessment (i.e. the AAT evaluation) and the presence of selective macrolinguistic impairments when the patients’ narrative abilities are at stack. This discrepancy suggests that standardized linguistic assessments may not be sensitive enough to capture an individual’s communicative performance. Indeed, in story description tasks speakers tend to be more fluid communicators and to make use of several linguistic skills in a communicatively oriented interaction. Consequently, discourse analysis allows controlling for the interaction among several processing levels (such as those between verb processing, argument structure generation, sentence production and inter-utterance integration). Furthermore, this analysis incorporates measures not included in traditional linguistic batteries, such as indexes of informativeness and of the individuals’ ability to generate cohesive and coherent ties across utterances. Therefore, the linguistic samples obtained in story description tasks may be long and complex enough to allow researchers and/or clinicians to evaluate a patient’s talkativeness in a more spontaneous context than standard linguistic tasks. All together, these data allow us to draw some important conclusions about the nature of the linguistic deficit in non-aphasic individuals after traumatic brain injury. Confirming previous results, the group of non-aphasic participants with TBI had normal lexical and grammatical skills and produced a normal amount of information content in their narratives (Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). However, this information was not correctly organized as they produced less lexical information units and more global coherence errors than the group of healthy participants. Our data suggest the possibility that the linguistic problems encountered by these patients may not be due to specific linguistic deficits as suggested by some authors (e.g. Peach & Schaude, 1986; Ellis & Peach, 2009) but to more generalized cognitive dysfunction (e.g. Coelho et al., 2005). Further research is required to explore the impact of these factors on the communicative impariment in people with traumatic brain injury. Conflict of interest No actual or potential conflicts of interest including any financial, personal or other relationships with other people or organizations that could inappropriately bias their work, is reported by any of the authors. Role of funding source Funding for this study was provided by IRCCS “E. Medea”. The IRCCS “E. Medea” had no further role in study design; in the col-
2909
lection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Acknowledgments The authors wish to thank the anonymous reviewers for their insightful comments to the first version of the paper. Appendix A. Litigio
Quarrel
E’ un un signore seduto in una poltrona che sta leggendo. . . / sta toccando. . . / leggendo un libro. . ./forse sua moglie. . . / e sua moglie gli gli dice di di leggere il libro. . . / di continuare a leggere il libro. . . / e e la moglie poi s’esce. . . / e lui quando vede la moglie che è uscita non legge più il libro / . . .e lui seduto nella poltrona dorme. . . / poi a un certo punto rientra la moglie / lui si alza . . ./ no no a un certo punto entra un’altra donna un’altra donna / lei entra / e lui parla con lei / poi l’abbraccia. . . / non so cosa fa. . . / l’abbraccia. . . / no forse la figlia. . . / l’abbraccia / e e lei era arrivata con una macchina che è andata addosso a un albero /
He is a a man sat on a chair who is reading . . . / is touching . . . / reading a book / perhaps his wife . . . / and his wife tells him him to to read the book / to keep on reading the book . . . / and and the wife then her goes out / and when he sees that his wife has left he stops reading the book / . . . and he sleeps sitting on the chair . . . / afterwards, at a certain moment, the wife comes back / he stands up . . . / no no at a certain point another woman another woman comes in / she comes in / and he talks to her / the he hugs her / I don’t know what he is doing / he hugs her . . . / no, perhaps the daughter . . . / he hugs her / and and she had arrived with a car who hit a tree
An example of narrative produced by one TBI participant. The story has been elicited in 102 s and comprised 125 units and 20 utterances with no phonetic errors. Therefore, he produced 125 phonologically well-formed words with a speech rate of 73.5 words per minute and a MLU of 6.3 words per utterance. Of the 125 words, 1 was a paragrammatic error (.8%), 3 were counted as semantic paraphasias (2.4%), 32 were repetitions (26%) and 7 were counted as lexical fillers (6%). Furthermore, of the 20 utterances, 4 showed omissions of morphosyntactic information (20%) and 16 of them were complete sentences (80%). When analyzing his macrolinguistic skills, the narrative description is characterized by 4 utterances abruptly interrupted (cohesive errors: 20%). Furthermore, 9 utterances showed violations of global coherence (global coherence errors: 45%). As for the functional analysis, only 65 of the 125 words produced in this description were lexical information units (%LIU = 52%). Finally, the analysis showed that the patient produced 8 of the 13 expected thematic units with a thematic density of .4. References Angeleri, R., Bosco, F. M., Zettin, M., Sacco, K., Colle, L., & Bara, B. G. (2008). Communicative impairment in traumatic brain injury: A complete pragmatic assessment. Brain and Language, 107, 229–245. Brookshire, B. L., Chapman, S. B., Song, J., & Levin, H. S. (2000). Cognitive and linguistic correlates of children’s discourse after closed head injury: A three year followup. Journal of the International Neuropsychological Society, 7, 741–750. Caplan, D. (1992). Language. Structure, processing and disorders. Cambridge, MA: MIT Press. Cannizzaro, M. S., & Coelho, C. A. (2002). Treatment of story grammar following traumatic brain injury: A pilot study. Brain Injury, 16, 1065–1073. Carlomagno, S., Giannotti, S., Vorano, L., & Marini, A. (2008). Discourse informativeness in TBI adults without aphasic symptoms. In The 46th Annual Meeting of the Academy of Aphasia Turku, Finland19th–21st October 2008. Chapman, S. B., Culhane, K. A., Levin, H. S., Harward, H., Mendelsohn, D., WwingCobbs, L., et al. (1992). Narrative discourse after closed head injury in children and adolescents. Brain and Language, 43, 42–65. Coelho, C. A. (2002). Story narratives of adults with closed head injury and nonbrain-injured adults: Influence of socioeconomic status, elicitation task, and executive functioning. Journal of Speech, Language, and Hearing Research, 45, 1232–1248. Coelho, C. A, Grela, B., Corso, M., Gamble, A., & Feinn, R. (2005). Microlinguistic deficits in the narrative discourse of adults with traumatic brain injury. Brain Injury, 19(13), 1139–1145.
2910
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
Davis, G. A., & Coehlo, C. A. (2004). Referential cohesion and logical coherence of narration after closed head injury. Brain and Language, 89, 508–523. Docking, K., Murdoch, B. E., & Jordan, F. M. (2000). Interpretation and comprehension of linguistic humor by adolescents with head injury: A group analysis. Brain Injury, 14, 89–108. Ehrlich, J. S. (1988). Selective characteristics of narrative discourse in head injured and normal adults. Journal of Communication Disorders, 21, 1–9. Ellis, C., & Peach, R. K. (2009). Sentence planning following traumatic brain injury. NeuroRehabilitation, 24(3), 255–266. Glosser, G. (1993). Discourse production patterns in neurologically impaired and aged populations. In H. H. Brownell, & Y. Joanette (Eds.), Narrative discourse in neurologically impaired and normal aging adults. San Diego, CA: Singular Publishing Group, Inc. Glosser, G., & Deser, T. (1990). Patterns of discourse production among neurological patients with fluent language disorders. Brain and Language, 40, 67–88. Haravon, A., Obler, L., & Sarno, M. (1994). A method for microanalysis of discourse in brain-damaged patients. In R. Bloom, L. Obler, S. De Santi, & J. Ehrlich (Eds.), Discourse analysis and applications: Studies in adult clinical populations (pp. 47–80). Hillsdale: Lawrence Erlbaum. Hartley, L. L., & Jensen, P. J. (1991). Narrative and procedural discourse after closed head injury. Brain Injury, 5, 267–285. Hartley, L. L., & Jensen, P. J. (1992). Three discourse profiles of closedhead-injury speakers: Theoretical and clinical implications. Brain Injury, 6, 271–282. Heaton, R. K., Chelune, G. J., Talley, J. L., Kay, G. G., & Curtiss, G. (1993). Wisconsin card sorting test manual, revised and expanded. Odessa: Psychological Assessment Resources. Huber, W., & Gleber, J. (1982). Linguistic and non-linguistic processing of narratives in aphasia. Brain and Language, 16, 1–18. Kertesz, A. (1982). Western Aphasia Battery. New York: Grune and Stratton. Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22, 1–37. Luzzatti, C., Willems, K., & DeBleser, R. (1991). Aachener Aphasie Test (Versione Italiana). Firenze: Organiszazioni Speciali. Marini, A., Boewe, A., Caltagirone, C., & Carlomagno, S. (2005). Age-related differences in the production of textual descriptions. Journal of Psycholinguistic Research, 34, 439–463. Marini, A., Carlomagno, S., Caltagirone, C., & Nocentini, U. (2005). The role played by the right hemisphere in the organization of complex textual structures. Brain and Language, 93, 46–54. Marini, A., Spoletini, I., Rubino, I. A., Ciuffa, M., Banfi, G., Siracusano, A., et al. (2008). The Language of Schizophrenia: An analysis of micro- and macrolin-
guistic abilities and their neuropsychological correlates. Schizophrenia Research, 105, 144–155. Marini, A., Andreetta, S., del Tin, S., & Carlomagno, S. (2011) A multi-level approach to the analysis of narrative language in Aphasia. Aphasiology, DOI:10.1080/02687038.2011.584690. McDonald, S., & Pearce, S. (1996). Clinical insight into pragmatic theory: Frontal lobe deficits and sarcasm. Brain and Language, 61, 88–104. Mentis, M., & Prutting, C. A. (1987). Cohesion in the discourse of normal and headinjured adults. Journal of Speech and Hearing Research, 30, 88–98. Nicholas, L., & Brookshire, R. (1993). A system for quantifying the informativeness and efficiency of the connected speech of adults with aphasia. Journal of Speech Hearing Research, 36, 338–350. Peach, R. K., & Schaude, B. A. (1986). Reformulating the notion of “preserved syntax following closed head injury”. In Paper Presented to the Annual Convention of the American Speech- Language-Hearing Association. Raven, J. C. (1938). Progressive matrices: A perceptual test of intelligence. London: H.K. Lewis. Reitan, R. M. (1992). Trail making test. Arizona: Reitan Neuropsychology Laboratory. Saffran, E. M., Berndt, R. S., & Schwartz, M. F. (1989). The quantitative analysis of agrammatic production: Procedure and data. Brain and Language, 37, 440–479. Shewan, C. M. (1988). The Shewan Spontaneous Language Analysis (SSLA) system for aphasic adults: Description, reliability and validity. Journal of Communication Disorders, 21, 103–138. Snow, P., Douglas, J., & Ponsford, J. (1998). Conversational discourse abilities following severe traumatic brain injury: A follow-up study. Brain Injury, 12, 911–935. Snow, P., & Douglas, J. (2000). Conceptual and methodological challenges in discourse assessment with TBI speakers: Towards an understanding. Brain Injury, 14(5), 397–415. Stierwalt, J. A. G., & Murray, L. L. (2002). Attention impairment following traumatic brain injury. Seminars in Speech and Language, 23(2), 129–138. Stout, C. L., Yorkston, K. M., & Pimental, J. D. (2000). Discourse production following mild, moderate and severe TBI: Comparison of two tasks. Journal of Medical Speech-Language Pathology, 8, 15–25. Thompson, C. K, Shapiro, L. P., Tait, M., Jacobs, B., & Schneider, S. L. (1996). Training wh-question production in agrammatic aphasia: Analysis of argument and adjunct movement. Brain and Language, 52, 175–228. Yorkston, K. M., & Beukelman, D. R. (1980). An analysis of connected speech samples of aphasic and normal speakers. Journal of Speech and Hearing Disorders, 45, 27–36.
Contents lists available at ScienceDirect
Neuropsychologia journal homepage: www.elsevier.com/locate/neuropsychologia
Narrative language in traumatic brain injury Andrea Marini a,b,∗ , Valentina Galetto c,d , Elisa Zampieri b , Lorenza Vorano e , Marina Zettin c,d , Sergio Carlomagno f a
Dipartimento di Scienze Umane, University of Udine, Udine, Italy IRCCS “E. Medea: La Nostra Famiglia”, San Vito al Tagliamento (Pn), Italy Centro Puzzle, Torino, Italy d Dipartimento di Psicologia, Università degli Studi di Torino, Italy e IMFR Gervasutta, Udine, Italy f Dipartimento di Psicologia, Università di Trieste, Italy b c
a r t i c l e
i n f o
Article history: Received 9 November 2010 Received in revised form 14 June 2011 Accepted 15 June 2011 Available online 22 June 2011 Keywords: Traumatic brain injury Narrative analysis Neurolinguistics Neuropsychology
a b s t r a c t Persons with traumatic brain injury (TBI) often show impaired linguistic and/or narrative abilities. The present study aimed to document the features of narrative discourse impairment in a group of adults with TBI. 14 severe TBI non-aphasic speakers (GCS < 8) in the phase of neurological stability and 14 neurologically intact participants were recruited for the experiment. Their cognitive, linguistic and narrative skills were thoroughly assessed. The group of non-aphasic individuals with TBI had normal lexical and grammatical skills. However, they produced narratives with increased errors of cohesion and coherence due to the frequent interruption of ongoing utterances, derailments and extraneous utterances that made their discourse vague and ambiguous. They produced a normal amount of thematic units (i.e. concepts) in their narratives. However, this information was not correctly organized at micro- and macrolinguistic levels of processing. A Principal Component Analysis showed that a single factor accounted for the production of global coherence errors, and the reduction of both propositional density at the utterance level and proportion of words that conveyed information. It is hypothesized that the linguistic deficits observed in the participants with TBI may reflect a deficit at the interface between cognitive and linguistic processing rather than a specific linguistic disturbance. © 2011 Elsevier Ltd. All rights reserved.
1. Introduction Individuals with severe traumatic brain injury (TBI) often show a variety of communicative difficulties. Indeed, problems in dealing with non-literal expressions (i.e. sarcasm, humor, proverbs, idioms, and indirect requests) have been frequently reported (Angeleri et al., 2008; Docking, Murdoch, & Jordan, 2000Docking, Murdoch, and Jordan, 2000; McDonald & Pearce, 1996). Furthermore, their discourse has been described as inefficient, impoverished or even confused (Davis & Coehlo, 2004; Hartley & Jensen, 1991, 1992). In the analysis of their linguistic performance an important distinction has been introduced between microlinguistic and macrolinguistic levels of discourse processing (e.g. Glosser & Deser, 1990). Microlinguistic abilities refer to within-sentence construction (i.e. lexical and morpho-syntactic aspects of language processing) analyzed in terms of lexical errors (e.g. phonological errors, verbal or semantic paraphasias, or use
∗ Corresponding author at: Università di Udine, Via Margreth, 3, 33100 Udine, Italy. Tel.: +39 335 5393224. E-mail address: [email protected] (A. Marini). 0028-3932/$ – see front matter © 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.neuropsychologia.2011.06.017
of indefinite terms) and/or syntactic organization (e.g. proportion of complete sentences, syntactic complexity) (Caplan, 1992). Macrolinguistic abilities relate to pragmatic and discourse-level aspects of language processing, i.e. those recruited in establishing cohesive and conceptual links among contiguous (cohesion and local coherence) or long-distant sentences/utterances (global coherence) and in formulating a mental model or gist of a story or procedure. Accordingly, macrolinguistic measures include the analysis of errors of cohesion, local and global coherence (e.g. presence of tangential utterances or extraneous propositional content). Several studies have shown potential dissociations between microlinguistic and macrolinguistic deficits in individuals with TBI. Nonetheless, Glosser and Deser (1990) reported that nine participants with TBI were impaired in lexical, syntactic and suprasentential organization. However, these persons were diagnosed with a form of fluent aphasia that may have hampered their microlinguistic processing. Different results come from other studies (e.g. Mentis & Prutting, 1987; Coelho, 2002). Indeed, when compared to a group of healthy participants, three individuals with TBI produced fewer cohesive ties in procedural descriptions but showed relatively preserved syntactic skills (Mentis & Prutting,
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
1987). In a similar vein, Coelho (2002) studied a cohort of 55 consecutive participants with TBI who were not aphasic on a story telling task. These individuals did not differ from a group of healthy participants in terms of sentence complexity and cohesive adequacy (i.e. proportion of complete cohesive ties) but introduced in their narratives more extraneous propositional content, suggesting problems in the organization of information at the between-sentence level. This shows that even cohesion and coherence may be dissociable components of discourse processing (see on this point Glosser, 1993). The reduced informative content in their discourse might contribute to the perception of impoverished and confused language. However, Ehrlich (1988) examined the Cookie Theft Picture descriptions from 10 non-aphasic adults with TBI with the Content Units method (CUs, Yorkston & Beukelman, 1980) and found them normal in the amount of CUs but inefficient. They uttered fewer CUs per minute than healthy participants. More recently, similar results have been reported by Stout, Yorkston, and Pimental (2000) and Carlomagno, Giannotti, Vorano, and Marini (2008) who analyzed the story-narratives from 94 and 10 non-aphasic participants with TBI, respectively. In the latter study, the persons with TBI showed normal microlinguistic processing but were found impaired on macrolinguistic measures. Interestingly, the occurrence of macrolinguistic errors correlated with the rating of language inaccuracy by naïve judges. These results further suggest that the impression of confused and impoverished language from non-aphasic individuals with TBI may depend on the reduced ability to organize information at the macrolinguistic level of processing rather than on difficulties in dealing with lexical and syntactic (i.e. microlinguistic) aspects of language production. The possibility of a problem in the global organization of information at the text level is supported by findings from studies focusing on story grammars. These refer to “the internal structure of stories which guide an individual’s comprehension and production of logical relationships, both temporal and causal, between agents and events” (Cannizzaro & Coelho, 2002, p. 1065). As such, story grammars rely on the interaction between language, memory and executive functions. Importantly, the ability to structure a story in terms of complete episodes seems altered in individuals with TBI (e.g. Brookshire, Chapman, Song, & Levin, 2000; Chapman et al., 1992; Davis & Coehlo, 2004). As for the distribution of information at the microlinguistic level, Coelho, Grela, Corso, Gamble, and Feinn (2005) analyzed the ability of a cohort of participants with TBI to insert multiple ideas into single sentences, i.e. evaluating the propositional density of their sentences. When compared to a group of healthy participants, the group of individuals with TBI produced significantly fewer propositions per T-unit (i.e. sentence). According to the authors, this suggests impairments of both micro- and macrolinguistic processes involved in the organization of semantic information in discourse. Furthermore, they proposed that such impairments “are symptomatic of generalized cognitive disruptions as opposed to specific linguistic deficits” (p. 1144). More recently, however, Ellis and Peach (2009) have examined the pausing patterns of TBI and healthy participants who had been administered sentence production (i.e. repetition and reading) tasks. They found increased pausing in the group of participants with TBI. Interestingly, their pausing pattern strongly correlated with a measure of sentence complexity. This led the authors to conclude that they may have specific deficits with online sentence processing even when normal on measures of syntactic complexity. Notably, in a preceding study by Peach and Schaude (1986) (quoted by Ellis & Peach, 2009), the narratives produced by 20 TBI patients were found normal on measures of syntactic complexity but contained more syntactic errors including word-order transposition, verb tense and agreement errors (see also Glosser & Deser, 1990).
2905
Overall, two major aspects of narrative discourse in non-aphasic individuals with TBI should be taken into account when interpreting the nature of their confused language. First, they seem to produce a normal amount of information content in their narratives although at reduced rate (Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). Second, this information may be poorly organized at both between-sentence level, as shown by the production of global coherence errors (Coelho, 2002; Carlomagno et al., 2008), and within-sentence level, as the results by Coelho et al. (2005) seem to indicate. However, a relevant question here is whether the microlinguistic deficit in the narratives produced by persons with TBI is due to specific linguistic problems (e.g. Ellis & Peach, 2009; Peach & Schaude, 1986) or to more generalized cognitive dysfunctions (e.g. Coelho et al., 2005). The present study further investigates the narrative skills of a group of non-aphasic individuals with TBI by analyzing in detail their narratives elicited in a single-picture and cartoon story description task. It was hypothesized that, in spite of preserved microlinguistic abilities and normal performance in terms of information content, these individuals would exhibit impaired organization of information at textual level producing several violations of global coherence. Furthermore, we expected to replicate the finding by Coelho et al. (2005) of reduced semantic content at the sentential level. Finally, we examined the relationship between linguistic and informative aspects of language production in TBI by performing a Principal Component Analysis. 2. Materials and methods 2.1. Participants Twenty-eight Italian-speaking participants were included in the study. They formed two groups balanced for age and level of formal education (Table 1). The experimental group was formed by 14 individuals with traumatic brain injury (TBI) (age – mean: 35.4; SD: 8.5; level of formal education – mean: 10.9; SD: 2.6). Their mean score on the Glasgow Coma Scale in the acute phase had been less than 8 (days of coma – mean: 32.4; SD: 18.4; range: 5–59) (see Table 1). In order to be included in the experimental group the patients had to be in the phase of neurological stability (months post-onset – mean: 68.5; SD: 38; range: 15–134) and normal to near-normal performance on the Aachener Aphasie Test (AAT, Italian version, Luzzatti, Willems, & DeBleser, 1991). The control group was formed by 14 neurologically healthy controls (HC) (Age – mean: 35.5; SD: 6.1; Level of formal education: 12.3; SD: 1.8) (Table 1). Inclusion criteria for admission in the control group included normal range performance on Raven’s progressive matrices (Raven, 1938) and normal performance on a series of neuropsychological tests (see Section 2.2.1 and Table 2). None of the participants had a previous history of psychiatric or neurological illness, learning disabilities nor hearing or visual loss. All participants released their written informed consent to participate to the study after all procedures had been fully explained. Approval for the study had previously been obtained from the local ethic committee. 2.2. Procedures 2.2.1. Neuropsychological assessment The cognitive profile of all participants was carefully evaluated. The neuropsychological assessment focused on those cognitive functions which are needed to complete the narrative task, i.e. phonological and semantic verbal fluency, phonological short-term memory (digit span), verbal learning (Rey’s 15-word Immediate Recall and Delayed Recall), executive functioning (Wisconsin Card Sorting Test, WCST perseverative and non-perseverative errors; Heaton, Chelune, Talley, Kay, &
Table 1 Means (and standard deviations) of demographic and clinical characteristics of the groups of traumatic brain injured (TBI) and healthy control (HC) participants. TBI
Age Formal education (years) Time after injury (months) Coma duration (days) GCS (score)
HC
Mean (SD)
(Range)
35.4 10.9 68.5 32.4 4.9
(18–50) 35.5 (8–13) 12.3 (15–134) (5–59) (3–8)
(8.5) (2.6) (38) (18.4) (1.7)
Mean
(SD) (Range) (6.1) (20–44) (1.8) (8–13) – – –
2906
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
Table 2 Means (and standard deviations) of the neuropsychological performance of the groups of traumatic brain injured (TBI) and healthy control (HC) participants.
TMT-A (s)* TMT-B (s)* Phonemic fluency Semantic fluency* WCST (pers. err.)* WCST (non-pers. err.)* Rey’s 15-word immediate recall* Rey’s 15-word delayed recall* Phonological short-term memory (digit span – forward)* Phonological short-term memory (digit span – backward)*
TBI
HC
82.6 (39.2) 198.6 (73.8) 24.1 (11.1) 12 (3.2) 47.4 (34.4) 18.4 (13.7) 27.7 (10) 3.1 (4.3) 5.1 (.9)
52.1 (11.1) 97.1 (28) 32.9 (8.6) 23.1 (3.9) .1 (.3) .2 (.4) 45.9 (6.7) 9.6 (1.9) 7 (1.1)
3.8 (.9)
5.6 (1.2)
GCS: Glasgow Coma Scale; TMT-A: Trail Making Test, Part A; TMT-B: Trail Making Test, Part B; WCST (pers. err): Wisconsin Card Sorting Test, perseverative errors; WCST (non-pers. err): Wisconsin Card Sorting Test, non-perseverative errors. * Significant group-related difference. The score on the phonological short-term memory was not available for the HC group.
Curtiss, 1993) and sustained and selective attention (Trail Making Test, Parts A and B, respectively; Reitan, 1992). Neuropsychological scores are shown in Table 2. 2.2.2. Assessment of narrative abilities The assessment of narrative abilities was performed using story telling. Each participant was asked to produce a set of narratives elicited with the help of one single picture depicting a shot of a story (the scene of a “Picnic”) and two cartoon stories with six pictures each presented on the same page (the stories of the “Flower Pot” and of a “Quarrel”). The single picture “Picnic” was taken from the Western Aphasia Battery (WAB; Kertesz, 1982). The two cartoon picture sequences have been used by Huber and Gleber (1982) and Nicholas and Brookshire (1993), respectively, for analyzing textual competence and discourse information content in brain damaged subjects. Each subject told all three stories. The order of presentation was counterbalanced across subjects. In order to avoid poor performance due to short-term memory limitations, the picture or cartoon story remained visible until the subject had finished his/her description. Each storytelling was tape-recorded and subsequently transcribed verbatim by three of the authors (AM, LV, VG); the transcription included phonological fillers, pauses, false starts and extraneous utterances. These transcriptions underwent quantitative, in-depth linguistic and textual analysis focusing on five main aspects of linguistic processing: productivity, speech rate, lexical and syntactic encoding, textual organization and informativeness (Marini, Andreetta, del Tin, & Carlomagno, 2011). Both simple values and ratios were considered to account for the difference across the narratives. An example of transcriptions and analyses of narratives from persons with TBI please is provided in Appendix A. Productivity measures included units, words, speech rate and mean length of utterance (MLU). The unit count included each word, non-word or syllabic false start uttered by the speaker. The total number of well-formed words with the exception of phonological fillers, phonemic paraphasias and phonetic errors was then computed. The number of words was used to obtain a measure of speech rate in terms of words per minute (Words/m ). For each story description, the total number of utterances was assessed following the criteria established in the Shewan Spontaneous Language Analysis System (Shewan, 1988). MLU was calculated by dividing the total number of words by the number of utterances. Lexical processing was assessed in a two-step analysis. At step one the analysis concerned the lexeme level of word processing (Levelt, Roelofs, & Meyer, 1999). The participants’ ability to retrieve phonologically well-formed words was assessed computing an index of phonological selection. This was obtained by dividing the number of words by the number of units (Marini, Carlomagno, Caltagirone, & Nocentini, 2005a). At step two the lemma level was assessed in terms of proportion of semantic and verbal paraphasias or paragrammatic errors in the sample. The former index was calculated by dividing the number of semantic and verbal paraphasias by the number of words (Haravon, Obler, & Sarno, 1994). The index of paragrammatic errors was calculated by dividing the number of paragrammatic errors by the number of words. These errors included misuse of bound morphemes and function words. The identification of grammatically complete sentences provided a measure of grammatical organization. An index was calculated by dividing the number of grammatical sentences by the number of utterances (Saffran, Berndt, & Schwartz, 1989; Thompson, Shapiro, Tait, Jacobs, & Schneider, 1996). A sentence was considered grammatically complete if all the arguments required by the verb were inserted correctly in the body of the sentence and if there were no omissions or substitutions of free or bound morphemes. As for discourse organization at intersentential level, the macrolinguistic measures included indexes of cohesion and global coherence errors. The index of
cohesive errors measured the extent to which each utterance of the narrative was structurally related to the preceding one. Cohesion errors were misuse of cohesive ties like anaphoric pronouns, errors in number and gender agreement between pronouns or noun phrases across utterances, misuse of either cohesive functionwords or semantically related content words, and abrupt interruptions of utterances (Haravon et al., 1994). The occurrence of cohesive errors was evaluated by dividing the total number of cohesive errors by the number of utterances produced (Marini, Boewe, Caltagirone, & Carlomagno, 2005b). With regard to global coherence, a ratio of global coherence errors was calculated by dividing the total number of tangential and extraneous utterances by the number of utterances that formed each description. The information content of each narrative was evaluated at three levels. The first one concerned the number of thematic units contained in the speech sample. A thematic unit was defined as a main idea or detail in the story that has been identified in a previous study (Marini, Carlomagno, et al., 2005). The number of thematic units produced by each subject was considered an index of the participants’ abilities to derive information from the picture stimuli. The second measure concerned the production of appropriate lexical information units (LIUs), i.e. those content and function words that were not only phonologically well-formed but also appropriate from a grammatical and pragmatic point of view (Marini, Boewe, et al., 2005; Marini, Carlomagno, et al., 2005). Therefore, all those words that were classified as semantic or verbal paraphasias, fillers, paragrammatic errors or forming tangential or extraneous utterances (i.e. utterances that were somehow deviating from the gist of the story) were excluded from the LIUs’ count. An index of lexical informativeness was then obtained by dividing the number of LIUs by the number of words. This ratio (%LIUs) has proved adequate in measuring effectiveness in encoding information in speech in several neurologically impaired populations and neuropsychiatric conditions (e.g. right brain damaged subjects, Marini, Carlomagno, et al., 2005; schizophrenic patients, Marini et al., 2008). Furthermore, in order to assess the participants’ ability to insert multiple ideas into single utterances, an index of “Thematic Density” was computed by dividing the number of thematic units produced by each participant for each story by the number of utterances that made up the speech-samples. This measure roughly reproduced the propositional complexity index used by Coelho et al. (2005) allowing us to examine the semantic complexity of utterances apart from sentence structure and grammaticality. The scoring procedure was performed independently by two raters and then compared. Acceptable interrater reliability was defined as k ≥ 0.80. After an appropriate training, the two raters, achieved this level prior to the beginning of the study. The residual differences were resolved through discussion. 2.3. Statistical analyses Group related differences in neuropsychological scores were assessed entering a Group (1. TBI; 2. C) X Task (phonemic and semantic fluency, TM-A, TM-B, and Rey’s 15-words immediate and delayed recall) MANOVA. As for the group-related differences on perseverative and non-perseverative errors on the WCST, a Mann–Whitney U test was performed. The narrative performance of the two groups of participants was analyzed performing a two-way ANOVA with repeated measures with group as between-subject factor and story as within-subject factor for each sample on twelve measures (words; speech rate; mean length of utterance; % phonological selection; % semantic/verbal paraphasias; % paragrammatic errors; % complete sentences; % cohesion errors; % global coherence errors; number of thematic units; % lexical information units; index of thematic density). The level of statistical significance was set at p < .004 after Bonferroni correction for muplitple comparisons. When significant, the story*group interaction will be reported. A correlation analysis was performed between neuropsychological and narrative scores in the group of participants with TBI. Finally, a Principal Component Analysis with Orthotran Varimax rotation was performed to identify factors explaining measures of informativeness and macrolinguistic processing.
3. Results 3.1. Cognitive and neuropsychological assessment Neuropsychological scores are shown in Table 2. The performance of the TBI group was significantly different from that of the group of healthy participants on all measures with the only exception of phonemic fluency. Namely, with respect to the control group they were slower on both TM-A ([F(1; 22) = 7.845; p = .010]) and TM-B ([F(1; 22) = 23.246; p = .000]) and they produced less words on the Semantic Fluency task ([F(1; 22) = 49.669; p = .000]). Furthermore, they recalled fewer items on Rey’s 15-word immediate ([F(1; 22) = 26.802; p = .000]) and delayed recall ([F(1; 22) = 22.595; p = .000]), respectively. The group-related difference
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
was also significant on the Digit Span test where the TBI group performed worse than the HC group on both forward and backward digit span ([F(1; 22) = 21.160; p = .000] and [F(1; 22) = 19.151; p = .000], respectively). As for the production of perseverative and non-perseverative errors on the WCST, the difference was significant in both cases: perseverative errors (Z = −4.498; p = .000); non-perseverative errors (Z = −4498; p = .000).
2907
3.2.3. Analysis of informativeness The mean values for each group on the three measures of informativeness are reported in Table 4. The group-related differences were not significant on thematic informativeness, i.e. the speech samples by the TBI subjects contained the same number of main ideas and details as the narratives from the healthy controls. However, the proportion of words that were counted as Lexical Information Units was reduced with respect to healthy controls ([F(1; 26) = 36.710; p = .000; partial 2 = .585]) indicating that relevant information was mixed with a number of unnecessary words, tangential speech or repetitions. Furthermore, the TBI participants also produced a significantly lower number of thematic units per utterance ([F(1; 26) = 35.251; p = .000; partial 2 = .576]). On the whole, the results of the information content analysis confirmed that narrative speech from non-aphasic TBI subjects contained a normal amount of information (see also Carlomagno et al., 2008; Ehrlich, 1988). This, however, was produced less efficiently as indicated by the reduced proportion of informative words (Carlomagno et al., 2008) and reduced semantic content per utterance.
3.2. Assessment of narrative abilities The results will be presented in three separate sections: a Microlinguistic analysis section, a Macrolinguistic analysis section and an Informative analysis section.
3.2.1. Microlinguistic analysis The mean values for each group on each microlinguistic measure are reported in Table 3. The two groups produced descriptions with a comparable number of words and MLU. However, the group of TBI participants produced narratives with significantly slower speech rate than the control group ([F(1; 26 = 13.684; p < .001; partial 2 = .345]). Both groups formulated descriptions with a similar index of phonological selection, semantic/verbal paraphasias and complete sentences. As for the production of paragrammatic errors, the group-related difference was significant ([F(1; 26) = 14.953; p < .001; partial 2 = .365]). However, it is noteworthy that the paragrammatic errors produced by the TBI participants were in any case very limited (see Table 3).
3.3. Correlations between neuropsychological scores and narrative performance in TBI participants A series of correlations were performed between those neuropsychological variable scores (i.e. non-perseverative and perseverative errors on the Wisconsin Card Sorting Test, Rey’s 15-word immediate recall and Rey’s 15-word delayed recall, Semantic Fluency, TMT-A and TMT-B) and those measures of the narrative production that were found altered in the TBI patients (i.e. speech rate, % cohesive errors, % global coherence errors, % lexical informativeness and the ratio of thematic density). The results failed to show any significant correlation between these scores.
3.2.2. Macrolinguistic analysis The mean values for each group on each macrolinguistic measure are reported in Table 4 together with data from the information content analysis. The TBI participants produced significantly more violations of both cohesion (F(1; 26 = 12.888); p < .001; partial 2 = .331]) and global coherence ([F(1; 26) = 31.134; p = .000; partial 2 = .553]) rules. However, the cohesion errors also included abrupt interruption of utterances (Haravon et al., 1994). When these errors were discarded from the computation, the other cohesive errors (e.g. misuse of anaphoric pronouns, errors in number and gender agreement between pronouns or noun phrases across utterances, misuse of either cohesive function-words or semantically related content words) did not differ in the two groups.
3.4. Principal component analysis of measures of informativeness and macrolinguistic accuracy In order to examine if separate factors corresponding to hypothesized components of micro- and macrolinguistic processing abilities could be empirically identified, data of each sample concerning measures of informativeness (% lexical information units and ratio of thematic density) entered a principal component analysis with Orthotran Varimax rotation together with the main measures of impaired macrolinguistic processing (% cohesion
Table 3 Results of the microlinguistic analysis for the groups of TBI and healthy control participants. Microlinguistic analysis
TBI
HC
Level of significance (p)
Effect size (partial 2 )
Words Speech rate* MLU % Phonological selection % Semantic paraphasias % Paragrammatic errors* % Complete sentences
82.5 (31.3) 94.7 (29.7) 5.6 (1.1) 99.2 (1) .8 (1.1) 1.4 (1.1) 57.9 (15.3)
80.9 (44.4) 129.5 (29.3) 6.9 (1.9) 99.6 (.8) .1 (.4) .2 (.6) 63.1 (23.1)
<.964 <.001 <.013 <.206 <.024 <.001 <.412
.000 .345 .213 .061 .181 .365 .026
*
When the group-related difference is significant after Bonferroni correction for multiple comparisons.
Table 4 Results of the analysis of the macrolinguistic and informative aspects of narrative production for the groups of TBI and healthy control participants. Macrolinguistic and informative analysis *
% Cohesive errors % Global coherence errors* % Lexical informativeness* Thematic informativeness Ratio of thematic density* *
TBI
HC
Level of significance (p)
Effect size (partial 2 )
3.9 (1.9) 22.1 (11.1) 64.3 (10.7) 6 (2) .4 (.2)
1.9 (.4) 3.8 (7.5) 84.5 (9.6) 7 (2) 1.2 (.6)
<. 001 .000 .000 <.037 .000
.331 .553 .585 .156 .576
When the group-related difference is significant after Bonferroni correction for multiple comparisons.
2908
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
errors and % global coherence errors) and speech rate. The latter was the only measure that could be assumed as a marker of microlinguistic deficit in the TBI group. Indeed, as mentioned above the % paragrammatic errors was higher in the participants with TBI but unlikely the small amount of these errors could explain the main features of their narrative speech. The principal component analysis showed that two factors could explain approximately 78% of variance (Bartlett’s 2 = 94.2, p < .0001). The first one, accounting for 50.5% of variance, included measures of % global coherence errors (−.93), % lexical information units (.93), and the ratio of thematic density (.83). The second factor, accounting for 27.5%, included the % cohesion errors (.80) and speech rate (−.79).
4. Discussion This study investigated cognitive, linguistic and narrative abilities of a group of severe TBI non-aphasic speakers while in the phase of neurological stability compared to a group of healthy participants. The neuropsychological assessment showed persisting difficulties in several areas of the cognitive domain. The analysis of narratives included microlinguistic, macrolinguistic and informative aspects of narrative language elicited in a cartoon-story description task. The main results indicate a significant deficit on the macrolinguistic and informative dimensions of processing in absence of important deficits of language processing on the withinsentence level. When evaluated on measures of verbal productivity and grammatical and lexical adequacy, the story narratives produced by the individuals with TBI were adequate. They had normal length (as measured in terms of uttered words and mean length of utterance), and showed no sign of phonological, morphological, semantic or syntactic deficits. Indeed, the only significant group-related difference concerned the production of paragrammatic errors and speech rate. The former, however, was limited to barely 1% of uttered words. Such a modest morpho-syntactic deficit unlikely accounts for the impoverished and confused language found in these patients. Interestingly, their narrative descriptions were produced with slower rate. Speech rate can be affected by several factors such as phonetic/phonological impairments, reduced lexical selection abilities or even macrolinguistic deficits. In our study, however, phonetic/phonological deficits can be ruled out, as the participants with TBI had normal phonological selection abilities. Similarly, problems in lexical access do not seem likely as shown by the absence of verbal or semantic paraphasias in their narratives. It remains open the possibility of a connection with macrolinguistic deficits. Indeed, the data discussed here showed reduced macrolinguistic competence in establishing the intersentential connections typical of a narrative structure. The narratives from the individuals with TBI were characterized by frequent interruptions of ongoing utterances (e.g. / the man is staring at . . . / the man is watching the dog /), and derailments (e.g. / It is a picnic / I like picnics / I have made several picninc in my life /) that made their discourse vague and ambiguous. The Principal Component Analysis showed that these macrolinguistic errors accounted not only for the reduced levels of cohesion and global coherence, but also for the reduced levels of speech rate and informativeness. Speech rate and Cohesion errors were explained by one single factor. Importantly, our evaluation of cohesion errors included the abrupt interruptions of ongoing utterances (Haravon et al., 1994). When interruption errors were discarded from the computation, the occurrence of cohesion errors, i.e. misuse of cohesive ties, was not different across the two groups. Therefore, it is likely that the reduced speech rate may have been determined by the frequent interruptions in the flow of thoughts rather than to specific linguistic deficits.
Another interesting finding concerns the organization of information in the narratives. The group of participants with TBI produced a normal amount of thematic units. This suggests that they could identify and introduce all expected thematic units (see also Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). However, they also exhibited reduced performance on two functional measures of informativeness. First, they produced a lower proportion of words that were scored as lexical information units (%LIUs). Second, their narratives contained fewer thematic units per utterance. The reduced %LIUs reflects a macrolinguistic deficit in coherently linking concepts in a story format by controlling what has been said, integrating the latter with incoming information, and inhibiting irrelevant behaviors or the production of tangential utterances. Indeed, words embedded in tangential utterances or forming extraneous propositional content were excluded from the LIUs count. However, the utterances produced by the participants with TBI had also less thematic density than those produced by the group of healthy participants. This finding is similar to that reported by Coelho et al. (2005) who interpreted it as a sign of generalized cognitive deficit influencing the organization of the semantic content of sentences. As mentioned in the Introduction section, Ellis and Peach (2009) suggested that the linguistic impairment found in persons with TBI may reflect specific linguistic deficits in online sentence planning. However, our data indicate that the participants TBI did not have relevant problems at the sentential level. Indeed, they produced the same amount of grammatically complete sentences as the healthy participants. Furthermore, a deficit in sentence planning could unlikely account for the production of errors of global coherence. Interestingly, in our study the principal component analysis indicated that a single factor accounted for the reduced %lexical information units, the reduced index of thematic density and the high % of global coherence errors. Taken together, all these symptoms may reflect a general cognitive problem in organizing the semantic content of the story rather than a specific linguistic deficit in coherently linking the propositions that form the narrative or in planning its sentences. This interpretation of the data is further supported by the neuropsychological profile of the participants. Indeed, the neuropsychological assessment showed important difficulties in several areas of the cognitive domain such as the ability to access lexical information stored in the mental lexicon by means of a semantic strategy (semantic verbal fluency), verbal learning skills (Rey’s 15-word Immediate Recall and Delayed Recall), short-term memory (forward and backward digit span), sustained and selective attention abilities (TM-A and TM-B), and executive functioning (perseverative and non-perseverative errors on the Wisconsin Card Sorting Test and phonemic fluency). The impairment in attention is among the most frequently reported sequelae of TBI (Stierwalt & Murray, 2002). This is a pervasive deficit that might affect several mental processes therefore contributing to other cognitive problems. As such, it may be involved in the observed difficulty in encoding and storing items to memory, in generating solution to problems and in language processing. In a similar vein, the anomalous performance on the WCST suggests that the participants with TBI may not be able to adequately select new strategies for solving problems and inhibit inappropriate responses. However, in our study none of these neuropsychological scores correlated with any of the linguistic measures mentioned so far (i.e. % lexical information units, index of thematic density, % global coherence errors). Even if this lack of correlation between linguistic and cognitive measures has been frequently reported in TBI, some studies have highlighted the potential relationship between executive functioning, memory and higher-level linguistic processing (e.g. Snow & Douglas, 2000; Snow, Douglas, & Ponsford, 1998). This apparent discrepancy can be explained in terms of a global cognitive impairment that becomes explicit when these patients fail to correctly organize the ideas they
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
intend to convey in a narrative. This poor organization of information may transcend the cognitive level and affect both macro- and microlinguistic processing even when the amount of information is normal. In other words, in a picture story description task as the one used in this experiment, the patient may extract all necessary information about the story but is then unable to organize it at both microlinguistic (as shown by the lowered propositional density of his/her utterances) and macrolinguistic (low %LIUs) levels. Such problems in the organization of information may therefore explain also the high production of macrolinguistic errors in terms of a compensatory strategy. Indeed, facing with this massive problem in organizing the information extracted by the pictures, the patient likely tries to express all these ideas randomly, without a clear organization. We therefore hypothesize that all these microand macrolinguistic symptoms may reflect a deficit in the interface between cognitive and linguistic processing as hypothesized by Glosser (1993). A further point that deserves discussion concerns the apparent discrepancy between the absence of linguistic deficits on the general linguistic assessment (i.e. the AAT evaluation) and the presence of selective macrolinguistic impairments when the patients’ narrative abilities are at stack. This discrepancy suggests that standardized linguistic assessments may not be sensitive enough to capture an individual’s communicative performance. Indeed, in story description tasks speakers tend to be more fluid communicators and to make use of several linguistic skills in a communicatively oriented interaction. Consequently, discourse analysis allows controlling for the interaction among several processing levels (such as those between verb processing, argument structure generation, sentence production and inter-utterance integration). Furthermore, this analysis incorporates measures not included in traditional linguistic batteries, such as indexes of informativeness and of the individuals’ ability to generate cohesive and coherent ties across utterances. Therefore, the linguistic samples obtained in story description tasks may be long and complex enough to allow researchers and/or clinicians to evaluate a patient’s talkativeness in a more spontaneous context than standard linguistic tasks. All together, these data allow us to draw some important conclusions about the nature of the linguistic deficit in non-aphasic individuals after traumatic brain injury. Confirming previous results, the group of non-aphasic participants with TBI had normal lexical and grammatical skills and produced a normal amount of information content in their narratives (Carlomagno et al., 2008; Ehrlich, 1988; Stout et al., 2000). However, this information was not correctly organized as they produced less lexical information units and more global coherence errors than the group of healthy participants. Our data suggest the possibility that the linguistic problems encountered by these patients may not be due to specific linguistic deficits as suggested by some authors (e.g. Peach & Schaude, 1986; Ellis & Peach, 2009) but to more generalized cognitive dysfunction (e.g. Coelho et al., 2005). Further research is required to explore the impact of these factors on the communicative impariment in people with traumatic brain injury. Conflict of interest No actual or potential conflicts of interest including any financial, personal or other relationships with other people or organizations that could inappropriately bias their work, is reported by any of the authors. Role of funding source Funding for this study was provided by IRCCS “E. Medea”. The IRCCS “E. Medea” had no further role in study design; in the col-
2909
lection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. Acknowledgments The authors wish to thank the anonymous reviewers for their insightful comments to the first version of the paper. Appendix A. Litigio
Quarrel
E’ un un signore seduto in una poltrona che sta leggendo. . . / sta toccando. . . / leggendo un libro. . ./forse sua moglie. . . / e sua moglie gli gli dice di di leggere il libro. . . / di continuare a leggere il libro. . . / e e la moglie poi s’esce. . . / e lui quando vede la moglie che è uscita non legge più il libro / . . .e lui seduto nella poltrona dorme. . . / poi a un certo punto rientra la moglie / lui si alza . . ./ no no a un certo punto entra un’altra donna un’altra donna / lei entra / e lui parla con lei / poi l’abbraccia. . . / non so cosa fa. . . / l’abbraccia. . . / no forse la figlia. . . / l’abbraccia / e e lei era arrivata con una macchina che è andata addosso a un albero /
He is a a man sat on a chair who is reading . . . / is touching . . . / reading a book / perhaps his wife . . . / and his wife tells him him to to read the book / to keep on reading the book . . . / and and the wife then her goes out / and when he sees that his wife has left he stops reading the book / . . . and he sleeps sitting on the chair . . . / afterwards, at a certain moment, the wife comes back / he stands up . . . / no no at a certain point another woman another woman comes in / she comes in / and he talks to her / the he hugs her / I don’t know what he is doing / he hugs her . . . / no, perhaps the daughter . . . / he hugs her / and and she had arrived with a car who hit a tree
An example of narrative produced by one TBI participant. The story has been elicited in 102 s and comprised 125 units and 20 utterances with no phonetic errors. Therefore, he produced 125 phonologically well-formed words with a speech rate of 73.5 words per minute and a MLU of 6.3 words per utterance. Of the 125 words, 1 was a paragrammatic error (.8%), 3 were counted as semantic paraphasias (2.4%), 32 were repetitions (26%) and 7 were counted as lexical fillers (6%). Furthermore, of the 20 utterances, 4 showed omissions of morphosyntactic information (20%) and 16 of them were complete sentences (80%). When analyzing his macrolinguistic skills, the narrative description is characterized by 4 utterances abruptly interrupted (cohesive errors: 20%). Furthermore, 9 utterances showed violations of global coherence (global coherence errors: 45%). As for the functional analysis, only 65 of the 125 words produced in this description were lexical information units (%LIU = 52%). Finally, the analysis showed that the patient produced 8 of the 13 expected thematic units with a thematic density of .4. References Angeleri, R., Bosco, F. M., Zettin, M., Sacco, K., Colle, L., & Bara, B. G. (2008). Communicative impairment in traumatic brain injury: A complete pragmatic assessment. Brain and Language, 107, 229–245. Brookshire, B. L., Chapman, S. B., Song, J., & Levin, H. S. (2000). Cognitive and linguistic correlates of children’s discourse after closed head injury: A three year followup. Journal of the International Neuropsychological Society, 7, 741–750. Caplan, D. (1992). Language. Structure, processing and disorders. Cambridge, MA: MIT Press. Cannizzaro, M. S., & Coelho, C. A. (2002). Treatment of story grammar following traumatic brain injury: A pilot study. Brain Injury, 16, 1065–1073. Carlomagno, S., Giannotti, S., Vorano, L., & Marini, A. (2008). Discourse informativeness in TBI adults without aphasic symptoms. In The 46th Annual Meeting of the Academy of Aphasia Turku, Finland19th–21st October 2008. Chapman, S. B., Culhane, K. A., Levin, H. S., Harward, H., Mendelsohn, D., WwingCobbs, L., et al. (1992). Narrative discourse after closed head injury in children and adolescents. Brain and Language, 43, 42–65. Coelho, C. A. (2002). Story narratives of adults with closed head injury and nonbrain-injured adults: Influence of socioeconomic status, elicitation task, and executive functioning. Journal of Speech, Language, and Hearing Research, 45, 1232–1248. Coelho, C. A, Grela, B., Corso, M., Gamble, A., & Feinn, R. (2005). Microlinguistic deficits in the narrative discourse of adults with traumatic brain injury. Brain Injury, 19(13), 1139–1145.
2910
A. Marini et al. / Neuropsychologia 49 (2011) 2904–2910
Davis, G. A., & Coehlo, C. A. (2004). Referential cohesion and logical coherence of narration after closed head injury. Brain and Language, 89, 508–523. Docking, K., Murdoch, B. E., & Jordan, F. M. (2000). Interpretation and comprehension of linguistic humor by adolescents with head injury: A group analysis. Brain Injury, 14, 89–108. Ehrlich, J. S. (1988). Selective characteristics of narrative discourse in head injured and normal adults. Journal of Communication Disorders, 21, 1–9. Ellis, C., & Peach, R. K. (2009). Sentence planning following traumatic brain injury. NeuroRehabilitation, 24(3), 255–266. Glosser, G. (1993). Discourse production patterns in neurologically impaired and aged populations. In H. H. Brownell, & Y. Joanette (Eds.), Narrative discourse in neurologically impaired and normal aging adults. San Diego, CA: Singular Publishing Group, Inc. Glosser, G., & Deser, T. (1990). Patterns of discourse production among neurological patients with fluent language disorders. Brain and Language, 40, 67–88. Haravon, A., Obler, L., & Sarno, M. (1994). A method for microanalysis of discourse in brain-damaged patients. In R. Bloom, L. Obler, S. De Santi, & J. Ehrlich (Eds.), Discourse analysis and applications: Studies in adult clinical populations (pp. 47–80). Hillsdale: Lawrence Erlbaum. Hartley, L. L., & Jensen, P. J. (1991). Narrative and procedural discourse after closed head injury. Brain Injury, 5, 267–285. Hartley, L. L., & Jensen, P. J. (1992). Three discourse profiles of closedhead-injury speakers: Theoretical and clinical implications. Brain Injury, 6, 271–282. Heaton, R. K., Chelune, G. J., Talley, J. L., Kay, G. G., & Curtiss, G. (1993). Wisconsin card sorting test manual, revised and expanded. Odessa: Psychological Assessment Resources. Huber, W., & Gleber, J. (1982). Linguistic and non-linguistic processing of narratives in aphasia. Brain and Language, 16, 1–18. Kertesz, A. (1982). Western Aphasia Battery. New York: Grune and Stratton. Levelt, W. J. M., Roelofs, A., & Meyer, A. S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22, 1–37. Luzzatti, C., Willems, K., & DeBleser, R. (1991). Aachener Aphasie Test (Versione Italiana). Firenze: Organiszazioni Speciali. Marini, A., Boewe, A., Caltagirone, C., & Carlomagno, S. (2005). Age-related differences in the production of textual descriptions. Journal of Psycholinguistic Research, 34, 439–463. Marini, A., Carlomagno, S., Caltagirone, C., & Nocentini, U. (2005). The role played by the right hemisphere in the organization of complex textual structures. Brain and Language, 93, 46–54. Marini, A., Spoletini, I., Rubino, I. A., Ciuffa, M., Banfi, G., Siracusano, A., et al. (2008). The Language of Schizophrenia: An analysis of micro- and macrolin-
guistic abilities and their neuropsychological correlates. Schizophrenia Research, 105, 144–155. Marini, A., Andreetta, S., del Tin, S., & Carlomagno, S. (2011) A multi-level approach to the analysis of narrative language in Aphasia. Aphasiology, DOI:10.1080/02687038.2011.584690. McDonald, S., & Pearce, S. (1996). Clinical insight into pragmatic theory: Frontal lobe deficits and sarcasm. Brain and Language, 61, 88–104. Mentis, M., & Prutting, C. A. (1987). Cohesion in the discourse of normal and headinjured adults. Journal of Speech and Hearing Research, 30, 88–98. Nicholas, L., & Brookshire, R. (1993). A system for quantifying the informativeness and efficiency of the connected speech of adults with aphasia. Journal of Speech Hearing Research, 36, 338–350. Peach, R. K., & Schaude, B. A. (1986). Reformulating the notion of “preserved syntax following closed head injury”. In Paper Presented to the Annual Convention of the American Speech- Language-Hearing Association. Raven, J. C. (1938). Progressive matrices: A perceptual test of intelligence. London: H.K. Lewis. Reitan, R. M. (1992). Trail making test. Arizona: Reitan Neuropsychology Laboratory. Saffran, E. M., Berndt, R. S., & Schwartz, M. F. (1989). The quantitative analysis of agrammatic production: Procedure and data. Brain and Language, 37, 440–479. Shewan, C. M. (1988). The Shewan Spontaneous Language Analysis (SSLA) system for aphasic adults: Description, reliability and validity. Journal of Communication Disorders, 21, 103–138. Snow, P., Douglas, J., & Ponsford, J. (1998). Conversational discourse abilities following severe traumatic brain injury: A follow-up study. Brain Injury, 12, 911–935. Snow, P., & Douglas, J. (2000). Conceptual and methodological challenges in discourse assessment with TBI speakers: Towards an understanding. Brain Injury, 14(5), 397–415. Stierwalt, J. A. G., & Murray, L. L. (2002). Attention impairment following traumatic brain injury. Seminars in Speech and Language, 23(2), 129–138. Stout, C. L., Yorkston, K. M., & Pimental, J. D. (2000). Discourse production following mild, moderate and severe TBI: Comparison of two tasks. Journal of Medical Speech-Language Pathology, 8, 15–25. Thompson, C. K, Shapiro, L. P., Tait, M., Jacobs, B., & Schneider, S. L. (1996). Training wh-question production in agrammatic aphasia: Analysis of argument and adjunct movement. Brain and Language, 52, 175–228. Yorkston, K. M., & Beukelman, D. R. (1980). An analysis of connected speech samples of aphasic and normal speakers. Journal of Speech and Hearing Disorders, 45, 27–36.